Effect of Head-To-Head Addition on Vinyl Acetate Propagation

Nov 21, 2014 - *E-mail [email protected]; Fax 7 96 45-22532 (A.N.N.)., ... p av, for the system, with the best-fit Arrhenius parameters of ln(A/...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/Macromolecules

Effect of Head-To-Head Addition on Vinyl Acetate Propagation Kinetics in Radical Polymerization Otlaatla Monyatsi,† Anatoly N. Nikitin,*,‡ and Robin A. Hutchinson*,† †

Department of Chemical Engineering, Dupuis Hall, Queen’s University, Kingston, Ontario K7L 3N6, Canada Institute on Laser and Information Technologies, Svyatoozerskaya 1, Shatura, Moscow Region 140700, Russia



S Supporting Information *

ABSTRACT: The radical propagation kinetics of vinyl acetate (VAc) has been studied by pulsed-laser polymerization coupled with analysis of the resulting polymer molar mass distributions. The significant increase in the apparent propagation rate coefficient observed with increasing pulse repetition rate is explained by the influence of the head-to-head defects formed during chain growth. Simulations that include head-to-head monomer addition are combined with an analysis of the experimental results to estimate the subsequent rate coefficients for reaction of the resultant radical as well as the propagation rate coefficient of normal head-to-tail addition. An analytical −1 expression is derived for an averaged rate coefficient, kav p , for the system, with the best-fit Arrhenius parameters of ln(A/L mol −1 s ) = 16.56 ± 0.35 and E/R= 2508 ± 108 K.



low relative to corresponding acrylate systems.8−10 In addition, Kattner and Buback11 have recently shown that no VAc midchain radicals could be detected using the powerful singlepulse PLP method coupled with time-resolved electron paramagnetic resonance measurement of radical concentrations (the SP-PLP-EPR technique). Kattner and Buback also highlighted another difficulty of using PLP to study VAc kinetics: while VAc radicals are very reactive, addition to the double bond of VAc monomer is slow. Ample evidence of the reduced reactivity of VAc (relative to acrylate or methacrylate monomers) to radical addition is found in the literature. In the 1940s, Mayo et al. investigated the copolymerization of VAc with eight representative monomers and found that the monomer is one of the least reactive of any common monomers toward radical attack.12 This low reactivity has been studied computationally and is attributed to the increased partial negative charge on the CH2 group resulting from the influence of the adjoining ester group that inhibits radical attack relative to other monomers.13 Additionally, the rate of VAc (co)polymerization is retarded in aromatic solvents such as toluene due to slow addition of monomer to the toluene radical formed after transfer to solvent.14−16 For effective PLP experimentation, the radicals formed by decomposition of the photoinitiator must be sufficiently active to attack the VAc double bond; if not, the efficiency of initiation suffers greatly, with some of the primary radicals formed only able to terminate existing growing chains rather than initiate new ones.11 As part of the current study, the

INTRODUCTION Mechanistic understanding of polymerization kinetics is a critical component required to guide the optimization of operating conditions for existing and new polymer grades, to discriminate between kinetic and physical effects during process development, to perform design and safety studies, and to understand and optimize transitions and other dynamics. These process development and modeling efforts are enabled by knowledge of the underlying polymerization mechanisms and accurate values of kinetic rate coefficients. A major advance has been the introduction of the PLP/MMD technique, in which pulsed-laser polymerization (PLP) is combined with analysis of the resulting polymer molecular mass distribution (MMD) to directly estimate kp, the chain growth rate coefficient for radical polymerization.1,2 However, the study of vinyl acetate (VAc) propagation kinetics by PLP has proven to be difficult. Originally believed due to high termination and transfer rates,3 more recent investigations suggest the difficulty arises from other kinetic features of the system. Junkers et al. recently have utilized pulse repetition rates up to 500 Hz laser to study VAc.4 Surprisingly, the apparent kp value (hereby referred to as kapp p ) increased by 33% as the pulse repetition rate (prr) was increased from 100 to 500 Hz. This behavior, not seen for “simple” monomers such as styrene and methacrylates,2 is an indicator that side reactions are confounding the kinetic analysis. Junkers et al. hypothesized that the influence of prr on VAc kpapp was a result of intramolecular transfer to polymer (backbiting), disrupting the relation between repetition rate and chain length, as has been found for acrylate systems.5−7 However, backbiting involving VAc during homopolymerization and copolymerization with ethylene is negligible or very © XXXX American Chemical Society

Received: October 22, 2014 Revised: November 11, 2014

A

dx.doi.org/10.1021/ma5021566 | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

where DP0 of the poly(VAc) is longer than 100 units but not for PLP experiments conducted at high repetition rates. Thus, we report here a systematic PLP/MMD investigation of the effect of prr on kapp p for bulk VAc between 25 and 65 °C. Three different photoinitiators are used to examine the possible confounding effect of ineffective initiation on the structure of the resulting MMD and kapp p estimates. The experimental study is accompanied by simulations that account for head-to-head addition as well as the reduced reactivity of the resulting inverted radical structure, with a comparison of simulation and experiment used to provide estimates for the VAc addition rate coefficients. Finally, an expression for an averaged propagation rate coefficient is derived that can be used to consider the influence of head-to-head insertion on continuously initiated polymerization systems.

influence of ineffective primary radicals is considered through simulation. Although photoinitiator choice has some influence on the structure of PLP-generated poly(VAc) MMDs (and thus determination of kapp p ), it is not sufficient to fully explain the apparent prr effect. Herein we present a plausible explanation based on a long-known feature of VAc systemshead-to-head monomer addition. The influence of head-to-head addition confounds the analysis of chain-growth kinetics, as shown by considering the possible modes of radical propagation associated with monomer structure shown in Scheme 1. For most monomers, Scheme 1. Propagation Reactions in Free Radical Homopolymerization, Accounting for the Possibility of Inverted Monomer Addition



EXPERIMENTAL SECTION

Vinyl acetate (VAc, Aldrich, 99+%), methyl methacrylate (MMA, Aldrich), 2,2-dimethoxyacetophenone (DMPA, Aldrich, 99%), dicumyl peroxide (DCP, Aldrich, 98%), and benzoin (Aldrich, ≥99.5%) were used as received. The experimental setup and procedure are similar to previous work.3,23,24 Bulk monomer solution of VAc (or MMA) with 5 mmol L−1 DMPA or benzoin photoinitiator were prepared; experiments with DCP were conducted with 90 mmol L−1 DCP, as suggested by previous literature.11 Samples were transferred into cuvettes and allowed to equilibrate to the desired temperature (±0.5 °C) before being exposed to the laser light generated by a Xantos XS500 laser operating at wavelength 351 nm, 3−15 ns pulse duration, and pulse energy of 3−8 mJ/pulse. The experiments were run to low monomer conversion ( 500 L mol−1 s−1. (iii)If the level of the head-to-head propagation is 10% (as for fluorinated α-olefins), the kav p value is close to kp11 only for the small region of kp21 in which kp21 = (kp11 − kp22)/2. For the unlikely case that kp21 > (kp11 − kp22)/2, the kav p value could be markedly higher than kp11 (curve 4). These calculations demonstrate the potential influence of headto-head addition on radical polymerization kinetics. For certain monomers such as VAc, the single kp value used widely to characterize the propagation must be considered as an averaged composite (kav p ) of four propagation rate coefficients, as given by expressions 3 and 4. The implications of this general finding will be explored after first using the variation of kapp p with prr to estimate individual rate coefficients for VAc propagation from the PLP/MMD results. Simulation of PLP/MMD Results. To simulate the experimental results obtained at different temperatures and prr, the basic model (initiation, propagation, termination, and

Table 3. Kinetic Parameters Used for Simulation of VAc Polymerization value at 50 °C (L mol−1 s−1)

kinetic parameters

Ea (kJ mol−1)

A (L mol−1 s−1)

kp11

23.7

6.31 × 107

9400

kp12

28.1

3.64 × 106

103

kp22

28.1

3.64 × 106

103

kp21

16.2

9.87 × 104

240

ktrM1 ktrM2

38.8 31.3

2.4 × 106 3.7 × 103

k1,1 t α1 α2 ic

9.0

3.3 × 1010

1.3 0.033 1.15 × 109 0.57 0.16 20

reference this work this work this work this work 35 this work 11 11 11 11

values, with the activation energy for kp11 chosen to be higher than the one for kav p to describe the experimental data. For kp12 the values of Ea and A are chosen to satisfy two conditions: the activation energy is chosen to be about 5 kJ higher than the one for kp11 in accordance with ref 20, with the pre-exponential fixed such that 0.01kp11 ≤ kp12 ≤ 0.02kp11, again in accordance with the literature evidence discussed previously. The value of kp22 is assumed to have the same Arrhenius parameters as for kp12; as shown previously (Figure 6), it was found by simulation that the value of kp22 does not influence significantly the calculated F

dx.doi.org/10.1021/ma5021566 | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

MMDs. The activation energy for kp21 was estimated in order to match the observed experimental activation energy of kav p of 20 kJ mol−1. Termination is modeled according to the composite model proposed by Smith et al.36 It is assumed that for both radicals Ri and Vi the termination mode is disproportionation and the termination rate coefficients ki,it are identical, as expressed by k ti , i = k t1,1i−α1 ,

i ≤ ic

k ti , i = k t1,1ic(−α1+ α2)i−α2 ,

(6a)

i ≥ ic

(6b)

In this expression, ic represents the chain length separating macromolecules with termination rate controlled by center-ofmass diffusion (i ≤ ic) from those for which segmental reorientation controls rate (i ≥ ic), with the specific parameters for VAc given in Table 3 taken from the literature.11 The rate coefficients of cross-termination are expressed using the geometric mean: k ti , j =

k ti , ik tj , j

(7)

The rate coefficient for chain transfer to monomer, ktrM1, for Ri is calculated from Arrhenius parameters found in the literature.35 For Vi chain transfer rate coefficient is assumed to be equal to (kp21/kp11)ktrM1. Rate coefficients for initiation ki1 and ki2 have been chosen to be equal kp11 and kp12, respectively. The model shown in Scheme 2 is enough to show the repetition rate dependence of kapp p . Nevertheless, the model is extended to account the fact that in experiments the increase of either pulse energy or initiator concentration results in disappearance of PLP structure in molar mass distributions.3 To explain this feature, it was assumed that each laser pulse creates specific radicals (Z) that behave as retarders; the concentration of these radicals is assumed to be proportional to the concentration of primary radicals produced by a pulse (ρ, in mol L−1). Then the effect of retardation is taken into account by the following reactions:

Figure 7. Polymer molar mass distributions simulated for vinyl acetate pulsed laser polymerization at 40.5 (a) and 50 °C (b) for different repetition rates in the presence of head-to-head addition. Calculated with σ (SEC dispersion parameter) = 0.04 for ρ = 3 × 10−7 (a) and 6 × 10−7 mol L−1 (b). Other kinetic parameters for calculations are given in Table 3.

kapp increases above kav p p only slightly and only when prr is increased well above 500 Hz; this feature of PLP experiments conducted at the low termination rate limit has been discussed before39 and is the reason why PLP conducted in this regime (very high prr) is not recommended for kp determination.40 (Note that our experimental MMDs, even at 500 Hz, have an easily observed secondary inflection point, indicating that these distributions were not measured at the high termination rate limit.39) For the simulations conducted with the head-to-head addition reaction included, the crossover point where kapp is p app equal to kav continues to p occurs at about 200 Hz, and kp increase toward the kp11 value, as the majority of radicals are not subjected to a head-to-head propagation event before the next pulse arrives. On the basis of these simulations, we conclude that the consideration of head-to-head addition is required to capture the extent of the variation in kapp p seen experimentally in both our data and that published by Junkers et al.4 We were only to match the experimental variation by setting kp11 (9400 L mol−1 −1 s−1 at 50 °C) to a much higher value than kav p (6750 L mol −1 app s ), with the PLP-determined kp values much closer to the latter value than the former. In addition and as presented in Table 4, the model does not capture the extent of the increase in activation energy of kapp p with the increase of pulse repetition rate observed experimentally (Table 2). Thus, the PLP-SEC data cannot be used to provide a reliable estimate of the true chain-end head-to-tail propagation rate coefficient (kp11) for VAc.

kz

R i + Z → Pi + R 0 k z2

Z + Z → P0

with kz = 1 × 109 L mol−1 s−1 and kz2 = 6 × 107 L mol−1 s−1. The model has been used to calculate MMDs at different prr for VAc polymerization at 40.5 °C (corresponding to the experiments of Junkers et al.4) and 50 °C, with the distributions plotted as Figure 7. The values of kapp p have been determined from such MMDs and compared to experimental results, as presented in Figure 8. To provide a comparison, MMDs have also been simulated assuming that VAc propagates solely through head-to-tail addition (kp11 = kav p , kp12 = kp22 = kp21 = 0) −1 −1 with the value of kav = 6750 L mol s ; all other kinetic p parameters used for this calculation were taken from Table 3. For both sets of simulation results shown in Figure 8, variation in kapp p with pulse repetition rate is observed. At low prr, the limiting value is slightly (